International Journal of Solids and Structures 94–95 (2016) 112–124

Contents lists available at ScienceDirect

International Journal of Solids and Structures

journal homepage: www.elsevier.com/locate/ijsolstr

2D mesoscale model for concrete based on the use of interface

element with a high aspect ratio
Eduardo A. Rodrigues a, Osvaldo L. Manzoli b,∗, Luís A.G. Bitencourt Jr. a,
Túlio N. Bittencourt a
a
University of São Paulo, Av. Prof. Luciano Gualberto, Trav. 3 n. 380, CEP 05508-010 São Paulo, SP, Brazil
b
São Paulo State University, UNESP/Bauru, Av. Eng. Luiz Edmundo C. Coube 14-01, CEP 17033-360 Bauru, SP, Brazil

a r t i c l e i n f o a b s t r a c t

Article history: The mesostructure of concrete plays a very important role in the process of initiation and propagation of
Received 1 October 2015 cracks. Microcracks tend to start in the Interfacial Transition Zone (ITZ) and propagate toward the mortar
Revised 7 April 2016
matrix until a macrocrack formation. Seeking to better understand the influence of the concrete meso-
Available online 12 May 2016
scopic structure, translated macroscopically in the form of loss of stiffness and energy dissipation, this
Keywords: work proposes a 2D mesoscale model in which the concrete is modeled as a heterogeneous three-phase
2D mesoscale analysis material composed of coarse aggregates, mortar matrix and ITZ. The coarse aggregates are generated
Interfacial transition zone from a grading curve and placed into the mortar matrix randomly. Interface solid finite elements with
Interface solid finite elements a high aspect ratio are used to represent the ITZ and the crack process based on a mesh fragmentation
Tension damage model technique. These interface elements present the same kinematics as the Continuum Strong Discontinuity
Concrete crack analysis Approach (CSDA), which allows the use of a continuum constitutive relation to describe their behavior.
Thus, an appropriate continuum tension damage model is adopted to describe the complex nonlinear
behavior of concrete due to the crack phenomenon. Initially, the proposed mesoscale approach is ap-
plied in uniaxial tensile tests to study the influence of the size, volume and distribution of the coarse
aggregates within the mortar matrix. Then, three-point bending beams are simulated in mesoscale and
the results compared with the experimental ones. The results showed that the proposed 2D mesoscale
model presents the same kinds of characteristics that real 3D concrete shows, considering the effects of
the mesostructure constituents.
© 2016 Elsevier Ltd. All rights reserved.

1. Introduction properties with the corresponding macroscopic response (López

Concrete is a composite material made of coarse aggregate,
sand (fine aggregate), cement and water. Nowadays, some addi-
tives are also included in the mixture. Therefore, concrete presents
a highly heterogeneous microstructure, which leads to a complex
structural behavior and cracking process, even for a simple exter-
nal load condition.
As is well known, the mesostructure affects the macroscopic re-
sponse of concrete and its internal structure plays a very important
role in the process of initiation and propagation of cracks. Hence,
numerical models in which concrete is assumed to be a homoge-
neous material are insufficient to rigorously establish the cause-
effect link between the geometrical characteristics and physical
∗
Corresponding author. Tel.: +55 11 310360 0 0.
E-mail address: omanzoli@feb.unesp.br, oslmanzoli@gmail.com (O.L. Manzoli).
et al., 2008).
Due to the recognized great influence of the internal structure
on the mechanical behavior of heterogeneous materials, allied to
the increased computational resources available nowadays, many
numerical models have been proposed to study the concrete ma-
terial at a mesoscopic level. Several models have been developed
based on lattice models (Grassl and Jirásek, 2010; Schlangen and
Garboczi, 1997; Schlangen and van Mier, 1992), in which the ma-
terial is represented by a discrete system of structural elements
(simple truss or beam elements) conveniently positioned, allowing
for the assignment of different mechanical properties for elements
failing in the mortar matrix, aggregates or ITZs. Other approaches
have been based on particle models (Bolander Jr. and Saito, 1998;
Cusatis et al., 2006; 2011; D’Addetta and Ramm, 2006; Zubelewicz
and Bažant, 1987). From the numerical analysis point of view,
these models are similar to the lattice models, also involving a dis-
crete system of structural elements, but they are computationally
more efficient due to the reduced number of unknowns. However,

http://dx.doi.org/10.1016/j.ijsolstr.2016.05.004
0020-7683/© 2016 Elsevier Ltd. All rights reserved.
 E.A. Rodrigues et al. / International Journal of Solids and Structures 94–95 (2016) 112–124
the calibration of the parameters that describe the interaction be-
tween particles is a very difficult task. Continuum FEM models,
characterized by the explicit representation of the mesoscopic con-
stituents, have also been proposed (Häfner et al., 2006; Unger and
Eckardt, 2011; Wang et al., 1999; Wriggers and Moftah, 2006). In
these models, when the goals of the research are focused on the
fracture process, discontinuities are introduced to the problem, us-
ing for example, special elements, such as zero-thickness interface
elements (Caballero et al., 20 07; 20 06; Carol et al., 2001; López
et al., 2008), or finite elements for capturing strong discontinuities
by means of elemental (E-FEM) (Oliver et al., 2015; Roubin et al.,
2015) or nodal enrichments (X-FEM) (Du et al., 2014).
In this paper, a 2D mesoscale model for concrete based on a
continuum FEM approach is proposed. The development of this
model involves two important tasks: generation of coarse aggre-
gates and discretization of the created mesoscopic model in 2D fi-
nite elements (FEs).
Regarding the generation of coarse aggregates, the existing al-
gorithms commonly take into account the shape, volume fraction
and distribution of the particles within the sample. The shape de-
pends on the aggregate type and, according to Wang et al. (1999), a
rounded shape can be used for gravel aggregates while the geom-
etry of crushed stone aggregates can be approximated by an an-
gular shape. For the sake of simplicity, regular geometries, such
as circular and elliptical shapes, are widely used (Bazant et al.,
1990; Eliáš and Stang, 2012; Grassl et al., 2012; Grassl and Pearce,
2010; Leite et al., 2004; Mier and Vliet, 2003; Wang et al., 2015).
Other researchers have also adopted arbitrary polygonal shapes
(López et al., 2008; Wang et al., 1999). Kim and Al-Rub (2011) have
investigated five different types of aggregate shapes in order to
study the influence of the shape on the material behavior. Re-
garding the aggregate size distribution, two techniques are often
used: grading curves and sieve analysis. Most researchers use the
Fuller curve when a grading curve is adopted (Bazant et al., 1990;
Carpinteri et al., 2004; Eliáš and Stang, 2012; Lilliu and van Mier,
2003; Schlangen and van Mier, 1992; Wittmann et al., 1993). Lastly,
the aggregates are placed within the sample by means of several
techniques, such as: take-and-place method (Häfner et al., 2006;
Wriggers and Moftah, 2006); divide-and-fill method (De Schut-
ter and Taerwe, 1993); Voronoi/Delaunay tessellations (Caballero
et al., 2006; Niknezhad et al., 2015); stochastic-heuristic algorithm
(Leite et al., 2004) and random particle drop method (Mier and
Vliet, 2003). More recently, image processing techniques, such as
scanning electron micrograph (Schlangen and Garboczi, 1997) or
X-ray computed tomography (Garboczi, 2002; Huang et al., 2015;
de Wolski et al., 2014) have been applied in order to obtain a
more realistic representation of the shape and distribution of the
particles.
The representation of the ITZ is the main challenge during the
FE mesh generation process. In addition, the application of an ap-
propriate strategy is essential due to the great influence of this
region on the initiation of cracks in concrete. As pointed out by
Monteiro et al. (1985), this region presents a very small thickness,
which requires a fine mesh along the ITZ. The so-called aligned
mesh approaches, in which the ITZ are explicitly represented, have
been used by several authors (Wang et al., 1999; Wittmann et al.,
1993; Wriggers and Moftah, 2006).
In this work an alternative model for modeling the concrete
behavior using a mesoscale approach is proposed. The concrete is
considered a three-phase material consisting of coarse aggregates,
mortar matrix, and ITZ. The main novelty of the approach pro-
posed is the use of standard low-order solid finite elements with
a high aspect ratio, developed recently by Manzoli et al. (2012), to
represent the ITZ and propagation through the mortar matrix by
means of a mesh fragmentation technique (Manzoli et al., 2016).
These elements are able to represent discontinuities by employing
113
an appropriate continuum constitutive model, as can be seen in
the recent successful applications for modeling of formation and
propagation of crack networks induced by shrinkage in drying soils
(Sánchez et al., 2014), cracks in quasi-brittle materials (Manzoli
et al., 2016), as well as for modeling of bond slip behavior of re-
inforcing bars embedded in concrete (Rodrigues et al., 2015). In
this work, the behavior of these elements (i.e. crack initiation and
propagation) is described by a tension damage model. Finally, this
constitutive model is integrated using an implicit-explicit (IMPL-
EX) integration scheme to avoid problems of convergence dur-
ing the analysis, commonly found in problems involving multiple
cracks (Oliver et al., 20 06; 20 08). In order to generate and place
the coarse aggregates into the mortar matrix, an efficient method
based on Monte Carlo’s simulation is employed. Thus, a grading
curve is used during the generation of aggregates, which are placed
into the mortar matrix randomly, so that no intersection between
aggregates is allowed, as proposed by Wriggers and Moftah (2006).
Moreover, for the sake of simplicity, regular geometry (octagon) for
the coarse aggregates was chosen, since the intention was only to
describe the new strategy developed for modeling concrete using
a mesoscale approach.
The remainder of this paper is outlined as follows. In Section 2,
the proposed mesoscale model for concrete is detailed. Attention is
given to the strategy used for modeling the coarse aggregates and
the presentation of the mesh fragmentation technique. The formu-
lation of the interface solid finite element and the continuum ten-
sion damage model used to describe its behavior are described in
Sections 3 and 4, respectively. In Section 5, numerical examples to
validate the technique are performed. Finally, the main conclusions
are drawn in Section 6.
2. Proposed mesoscale model for concrete
In the proposed mesoscale model for concrete, the coarse ag-
gregate particles, mortar matrix and ITZ are modeled in 2D as
separate constituents. Initially, the coarse aggregate particles are
generated and placed within the domain of the sample using
the “take-and-place” method proposed by Wriggers and Moftah
(2006). Then, the FE mesh is generated and a mesh fragmentation
procedure is applied to represent the ITZ and to define all the po-
tential crack paths. The model is completed by adopting a tension
damage model to describe the initiation and propagation of cracks
under quasi-static loading.
The main characteristics of this mesoscale model are: (i) the
ITZ is modeled explicitly by the use of interface elements with a
high aspect ratio and its mechanical properties can be defined sep-
arately; (ii) the use of fine meshes is not necessary to describe
the ITZ; (iii) multiple arbitrary cracks can be simulated without
the need of tracking algorithms; and (iv) the model is developed
entirely in the continuum framework, composed of standard finite
elements, usually available in FEM codes, and continuum constitu-
tive models.
2.1. Modeling of coarse aggregate
For the sake of simplicity, coarse aggregate particles are
geometrically represented by regular octagons. The efficient “take-
and-place” method proposed by Wriggers and Moftah (2006) is
employed to generate the aggregate sizes from a grading curve,
and to locate each particle in the mortar matrix randomly, so that
intersection between aggregates is avoided. In addition, the grad-
ing curve may be constructed based on an experimental sieving
process. Thus, for an interval [ds , ds+1 ] defined by two sequential
sieve opening sizes, ds and ds+1 , the passing percentage is defined
114 E.A. Rodrigues et al. / International Journal of Solids and Structures 94–95 (2016) 112–124
Fig. 1. Mesh fragmentation technique: (a) generation of the standard FE mesh to
be fragmented, (b) separation of the finite elements by introducing gaps in between
them, and (c) insertion of pairs of interface elements.
according to the equation:
P (ds+1 ) − P (ds )
Vp [ds , ds+1 ] = .Vagg
P (dmax ) − P (dmin )
where dmin and dmax are the minimum and maximum particle
sizes, respectively, and Vagg is the total volume fraction of coarse
aggregate particles assumed for the concrete. It is important to
note that, within this interval, the diameter of the aggregate can
be defined randomly, assuming values between ds and ds+1 .
Alternatively, the grading of aggregate particles may be defined
by the Fuller curve, which gives an optimal packing of aggregates.
The Fuller curve can be expressed by the following equation:
 n
di
Pi = 100
dmax
where Pi and di are the percent passing and opening size of the
ith sieve, respectively, dmax is the maximum particle size and the
exponent n assumes value between 0.45 and 0.70.
2.2. The fracture representation in the mortar matrix and ITZ
The most important step in the development of the approach
proposed is the application of a mesh fragmentation technique.
This technique is based on the use of standard low-order solid fi-
nite elements with a high aspect ratio to describe discontinuity by
means of a continuum constitutive relation, as recently proposed
by Manzoli et al. (2012). Thus, these interface elements are used
in this work to represent the ITZ and to define all potential crack
paths.
a b
Fig. 2. Mesh fragmentation technique for a sample with one embedded aggregate: (a) original mesh, (b) introduction of gaps between all the matrix–matrix and matrix-
aggregate (ITZ) elements, and (c) insertion of interface elements.
Fig. 1 illustrates the main steps of the mesh fragmentation tech-
nique. First, a regular (initial) FE mesh is generated (Fig. 1(a)).
Then, the finite elements are separated by duplicating the com-
mon nodes and reducing their sizes, so that very narrow gaps are
created between the modified regular elements (Fig. 1(b)). Finally,
to fill these very thin gaps, pairs of three-node triangular elements
with a high aspect ratio are inserted (Fig. 1(c)).
Note that for this simple example, in which the original mesh
is composed of four three-node triangular finite elements and five
nodes (see Fig. 1(a)), the fragmentation procedure requires the ad-
dition of seven extras nodes (p, q, r, s, t, u and v), as can be seen
in Fig. 1(b). Hence, eight interface elements are inserted: IE1 =
{m, l, p}, IE2 = { p, l, q}, IE3 = { p, n, r}, IE4 = {r, n, s}, IE5 = {r, o, t },
IE6 = {t, o, u}, IE7 = {t, v, m} and IE8 = {m, v, k}, as can be observed
in Fig. 1(c). Also note that the interface elements are defined based
on the nodes of the modified regular elements and their thick-
ness are with an exaggerated scale factor for clarity. Usually, a very
small thickness (approximately 0.01 mm) is adopted.
Concrete with normal strength is assumed for the problems ad-
(1)
dressed in this paper. Hence, only the FE mesh of the mortar ma-
trix and ITZ are fragmented, as illustrated in the Fig. 2 for a sample
with one coarse aggregate particle. Note in this figure that distinct
material models can be easily used to describe the failure behavior
of the ITZ and the fracture process in the mortar matrix.
3. Interface solid finite element
The interface solid finite element proposed by Manzoli et al.
(2012) corresponds to a standard three-node triangular finite ele-
ment with a height h tending to zero (h → 0), as depicted in Fig. 3.
(2) Thus, to understand the behavior of this element, it is appropri-
ate to divide the FE approximation of the strain field, ε, as follows
(Manzoli et al., 2012):
1
ε = ε˜ + εˆ = ε˜ + (n  [[u]] )S (3)
h  
εˆ
where εˆ collects the components that depend on the height h, and
ε˜ contains the rest of the components. The notation (•)S refers to
the symmetric part of (•),  denotes a dyadic product, n is the
unit vector normal to the element base and [[u]] is a vector that
contains the components of the relative displacement between the
node (1) and its projection on the element base (1 ).
Note that, in the Eq. (3), when h tends to zero, the component ε˜
remains bounded, while the εˆ component is no longer bounded. As
a consequence, the element strains are related almost exclusively
c
 E.A. Rodrigues et al. / International Journal of Solids and Structures 94–95 (2016) 112–124
Fig. 3. Triangular interface finite element with a high aspect ratio.
to the relative displacement [[u]], which becomes the measure of
a displacement discontinuity (strong discontinuity), and the struc-
ture of the strain field in Eq. (3) corresponds to the typical kine-
matics of the Continuum Strong Discontinuity Approach (CSDA).
Details regarding the equivalence between the strain field of the
interface finite elements and the strong discontinuity kinematics
can be found in Manzoli et al. (2012).
Therefore, based on the same concepts of the CSDA, it can
be stated that bounded stress can be obtained from unbounded
strains by means of a continuum constitutive relation. In the fol-
lowing section, the continuum tension damage model used to de-
scribe the behavior of the interface elements is described.
4. Tension damage model
A constitutive model based on the Continuum Damage Mechan-
ics Theory (CDMT) is used to describe the complex phenomenon of
crack initiation and propagation through the interface solid finite
elements. The formulation of this model is based on the effective
stress field associated with a scalar damage variable, which varies
from d = 0 (undamaged material) to d = 1 (totally damaged mate-
rial). The damage criterion is based on the component of the stress
component normal to the base of the element, σ̄nn . The damage
variable degrades all the components of the effective stress tensor
(σ̄ ) to calculate the nominal stress tensor (σ ). Thus, the proposed
model is able to describe the crack propagation process controlled
by a prevalent mode I.
Table 1 summarizes all the equations of the tension damage
model, where C is the fourth order elastic tensor, ε is the strain
tensor, r is the strain-like internal variable, ft is the material ten-
sile strength, h is the element’s thickness and A is the softening
parameter, which is related to the fracture energy Gf . More details
regarding the relation between the continuum and discrete consti-
tutive equations, in the context of CSDA, can be found in Manzoli
et al. (2016).
4.1. IMPL-EX Integration scheme
To increase the stability and the robustness of the solution
strategy, the IMPL-EX integration scheme proposed by Oliver et al.
Table 1
Equations of the tension damage model.
Effective stresses σ̄ = C : ε
Equivalent stress τ̄ = σ̄nn
Damage criterion φ̄ = τ̄ − r
Evolution law for the strain-type variable r (t ) = max[σ̄nn (s ), ft ]
s∈[
0,t ]
Ah 1−
Damage evolution d (r ) = 1 − frt e
Constitutive relation σ = (1 − d )σ̄
115
(2006) is used to integrate the tension damage model. Recently,
this scheme has been used in several types of applications, for ex-
ample for elastoplasticity problems (Oliver et al., 2008; Prazeres
et al., 2016) and those described by damage mechanics (Oliver
et al., 2008). The idea behind of this strategy is to combine the
stability offered by explicit methods with the precision of the im-
plicit ones. The result is the guarantee of convergence with only
one iteration per load step and the positive definite character of
the algorithmic tangent operator. Despite its large advantages from
computational point of view, there is an error associated to the use
of this integration scheme, which can be reduced (or controlled) by
decreasing the load step length.
Fig. 4 shows a flowchart detailing the integration of the ten-
sion damage model at the current pseudo-time tn+1 , which can be
briefly described by the following steps:
1. Evaluate the effective stress tensor σ̄ (n+1 ) ;
2. Implicit computation of the internal strain-like variable r(n+1 ) ;
3. Computes an explicit linear extrapolation of the internal strain-
like variable in terms of the implicit values obtained in the pre-
vious time steps, r˜(n+1 ) r˜(n ) , r˜(n−1 ) ;
4. Evaluate the damage variable as function of the extrapolated
strain-like variable d(n+1 ) r˜(n+1 ) ;
5. Compute the nominal stress tensor σ˜ (n+1 ) ;
tang
6. Evaluate the effective algorithmic tangent operator C˜ (n+1 ) .
Note that the integration procedure is performed in a closed
form, in which, from the given current strain tensor, εn+1 , the
stress tensor, σ˜ n+1 , is evaluated explicitly and used to fulfill the
balance equation and to compute the effective algorithmic tangent
operator. The first two steps correspond to the implicit stage and
the others to the explicit stage.
5. Numerical examples
The proposed mesoscale model for concrete is applied in two
numerical examples. First, in order to study the effects of the ag-
gregate parameters (size, volume fraction and distribution) on the
material response, several uniaxial tension tests are performed. Af-
ter, three-point bending beams are numerically analyzed and the
results are compared with the experimental ones to investigate the
capability of the strategy to simulate structural members.
Concrete with normal strength is assumed for all the examples.
Thus, aggregate particles are described by linear elastic behavior
and the material nonlinear response is given exclusively by the
cracks represented by the mesh fragmentation technique. In ad-
dition, the tensile strength for the ITZs is assumed to be 50% of
the tensile strength of the mortar matrix. This assumption induces
r crack initiation along the ITZs. A thickness of 0.01 mm was adopted
ft
for the interface finite elements used to represent both the ITZs
and crack paths in the mortar matrix.
116 E.A. Rodrigues et al. / International Journal of Solids and Structures 94–95 (2016) 112–124
Fig. 4. Flowchart of the IMPL-EX integration scheme for the tension damage model.
5.1. Uniaxial tension tests
Several specimens with square cross section (150 mm ×
150mm) under uniaxial tension load are analyzed. The analyses are
performed in plane stress condition with an out-of-plane thickness
of 150 mm. Fig. 5(a) shows the test setup, including the bound-
ary conditions. As can be seen, the specimen is loaded by impos-
ing an increasing prescribed vertical displacement along the upper
face, while vertical displacements with value equal to zero are pre-
scribed to all nodes of the lower face. The FE mesh after fragmen-
tation procedure is depicted in Fig. 5(b). Note that interface solid
elements are inserted between all the elements of the mortar ma-
trix and between this region and the aggregate particles. However,
as the ITZ is the weakest part of the material, different mechanical
properties are adopted to describe the crack initiation along these
thin interfaces. Following the recommendations of some authors
(Huang et al., 2015; Kim and Al-Rub, 2011), the elastic modulus
adopted for the ITZ is a little smaller than the mortar matrix. It is
also important to mention that this small reduction has very little
influence on the mechanical response of the interface element, as
can be seen in Manzoli et al. (2016). The parameters adopted for
each phase of the material are summarized in Table 2.
In the following sections, the effects of size, volume fraction
and distribution of coarse aggregate particles on the failure pro-
cess of concrete under tension are investigated. For all these cases
studied, the particles are generated and placed randomly within
the specimen using the method described in Section 2.1. For all of
them, Fuller’s grading curve is employed with n = 0.45.
5.1.1. Effect of aggregate size
As a first case of study, four specimens were simulated for max-
imum aggregate diameter of 10, 20, 30 and 40 mm, maintaining a
constant aggregate volume fraction of 40% and minimum aggregate
diameter of 4.75 mm.
The vertical normal stress and crack propagation process for
the case with maximum aggregate diameter of 30 mm are illus-
trated in Fig. 6, which can be better understood with the aid of
the graph plotted in Fig. 7. Thus, a first load stage can be character-
ized by a linear elastic response, until about 60% of the peak-load
(Fig. 7– stage I), and where no damage is observed in the sample
 E.A. Rodrigues et al. / International Journal of Solids and Structures 94–95 (2016) 112–124 117

a b

Fig. 5. Specimen under tension load: (a) geometry and boundary condition and (b) close-up showing the details of the fragmented mesh.

Table 2
Material parameters for the uniaxial tension tests.

Materials Elastic modulus Poisson’s ratio Fracture energy Tensile strength

Coarse aggregate Eagg = 37.0 GPa νagg = 0.2 – –

Mortar matrix Em = 20.0 GPa νm = 0 . 2 – –
Matrix–matrix interface Emi = 20.0 GPa νmi = 0 G fmi = 0.03 N/mm ftmi = 2.6 MPa
ITZ Eitz = 18.0 GPa νitz = 0 G fitz = 0.02 N/mm ftitz = 1.3 MPa

a b

Fig. 6. Vertical normal stress and crack pattern for the case with maximum aggregate diameter of 30 mm: (a) stage I: non-damaged material (scaling factor of 300), (b)
stage II: microcracks at the ITZs (scaling factor of 300) and (c) stage III: macrocrack formation due to the microcracks coalescence (scaling factor of 50).

(Fig. 6(a)). Above this load, the damage starts and a number of mi-
crocracks are simultaneously formed, predominantly perpendicular
to the load direction at the ITZs, as can be seen in Fig. 6(b). In the
structural response, this distributed damage configuration repre-
sents the transition from the linear to nonlinear regime before the
peak-load (Fig. 7– stage II). Then, with increasing load, some mi-
crocracks propagate through the mortar matrix, while at the same
time, others are being unloaded or even closed. At this final stage,
the microcracks coalesce resulting in a predominant crack located
at a narrow region of the specimen (Fig. 6(c)), corresponding to the
softening behavior of the structural curve (Fig. 7– stage III).
The bridging mechanisms between cracks observed in experi-
ments (van Mier, 1991; 1997) could also be observed in the nu-
merical results (see detail of the Fig. 6(c)), where the aggregate or
part of the matrix is superimposed by cracking (interlocking aggre-
gate or overlapping crack segments), transmitting loading between
their faces and contributing to residual force exhibited in the curve
depicted in Fig. 7. Moreover, the superposition of cracks may be
responsible for a significant increase of the fracture energy, since
the bridging mechanism may remain active during the macrocrack
propagation and can contribute significantly to the deviation of the
macrocrack formation, consequently increasing the fracture surface
area, as already reported in previous works (Landis and Bolander,
2009; van Mier and Nooru-Mohamed, 1990).
Final crack patterns for the four cases studied are shown in
Fig. 8. As can be seen in this figure, the crack pattern depends
118 E.A. Rodrigues et al. / International Journal of Solids and Structures 94–95 (2016) 112–124

16

14
Stage I - Elastic regime
12
Stage II - Hardening
10

Load (kN)
Stage III - Softening
8

0
0 0.05 0.1 0.15 0.2
Displacement (mm)
Fig. 7. Different stages of load-displacement response for the specimen with maximum aggregate diameter of 30 mm.

a b 16

14 dmax=10mm dmax=20mm

12
Load (kN) dmax=30mm dmax=40mm
10

c d 4

0
0 0.05 0.1 0.15 0.2
Displacement (mm)
Fig. 9. Load-displacement curves for the four specimens with maximum aggregate
diameter of 10, 20, 30 and 40 mm.

Fig. 8. Final crack patterns obtained for different maximum aggregate diameters:
(a) d = 10 mm (showing the IE), (b) d = 20 mm (showing the IE), (c) d = 30 mm
(without showing the IE) and (d) d = 40 mm (without showing the IE). (figures with
a scaling factor of 50).
on the aggregate size. This effect becomes clear by observing the
graphs in Fig. 9, in which it can be noted that increasing the aggre-
gate size leads to a significant peak resistance reduction and a pro-
nounced variation of the amount of dissipated energy for the same
aggregate volume fraction. The less pronounced softening branch
shown can be explained by the bridging or other form of crack face
interaction that tends to increase the residual strength and, conse-
quently, the area under the curve and so also the dissipated en-
ergy. These same qualitative and quantitative behaviors have been
observed by some authors in numerical studies (Du et al., 2014;
Kim and Al-Rub, 2011; López et al., 2008) and experimental inves-
tigations, for samples subjected to direct tension (Hordijk, 1992) or
three-point bending tests (Zhang et al., 2010).
To clearly illustrate the detailed crack initiation and propagation
processes, the deformed meshes of the specimens are showed in
Fig. 6 by adopting different scaling factors. Furthermore, the inter-
face solid elements may or may not be shown in the post-process,
as illustrated in Fig. 8(a) and (b), showing the deformed interface
elements, or in Fig. 8(c) and (d), without showing them. However,
in the remainder of this work the interface elements are no longer
shown since the crack paths are clearer without them.
5.1.2. Effect of aggregate volume fraction
Specimens with coarse aggregate volume fractions of 20, 30,
40 and 50% were numerically simulated for the same aggregate
gradation (aggregate diameter varying the between dmin = 5 mm
and dmax = 10 mm). The results obtained in terms of crack patterns
are depicted in Fig. 10 and their corresponding load-displacement
curves are plotted in Fig. 11. As shown, the crack paths for the
four cases are distinct. A lower aggregate volume fraction leads to
a higher peak-load. Similar results were obtained in previous nu-
merical studies (Kim and Al-Rub, 2011; López et al., 2008; Wang
et al., 2015).
Since the behavior of the coarse aggregates is described by a
linear elastic response, the results obtained consolidate the fact
that the ITZ plays a very important role in the concrete structural
behavior under tension load. As reported by Landis and Bolan-
der (2009), a high percentage of aggregate can induce coupling
between interfaces, forming paths conducive to the formation of
crack (percolation of interfaces zones) which can contribute to re-
ducing the tensile strength of concrete.
5.1.3. Effect of aggregate distribution
Four random generations of aggregates are performed to study
the effect of aggregate distribution for a constant volume fraction
 E.A. Rodrigues et al. / International Journal of Solids and Structures 94–95 (2016) 112–124 119

a b a b

c d c d

Fig. 10. Final crack patterns for different aggregate volume fractions: (a) Vagg = 20%, Fig. 12. Final crack patterns for different random realizations of coarse aggregate:
(b) Vagg = 30%, (c) Vagg = 40% and (d) Vagg = 50% (scaling factor of 50). (a) realization 1, (b) realization 2, (c) realization 3 and (d) realization 4 (scaling
factor of 50).

of aggregate equal to 40% and for the same aggregate gradation
with minimum and maximum diameter of aggregate equal to
dmin = 5 mm and dmax = 10 mm, respectively. As expected, the
final crack patterns obtained for the four cases are distinct, as
can be seen in Fig. 12. Furthermore, it is possible to note in the
load-displacement curves illustrated in Fig. 13 that the randomly
aggregate distribution affects only the post-peak structural re-
sponse of the numerical simulations. Thus, it is reasonable to state
that for this case study, the aggregate distribution has a negligible
influence on the mechanical properties of concrete until the
peak-load. Furthermore, the slightly different softening branches
can be justified by the distinct crack paths obtained (see Fig. 12).
According to Tejchman and Bobiński (2013), this may explain the
difficulty in finding a representative volume for random heteroge-
neous material like concrete, for which the process zone of strain
localization occurs at the weakest zone, which is dependent on
their constituent material arrangements.
The results obtained in these numerical simulations are in good
agreement with those obtained by other researchers (Kim and Al-
Rub, 2011; Wang et al., 2015).
5.2. Three-point bending tests
A series of three-point bending tests were experimentally per-
formed by Grégoire et al. (2013). In that research, notched and
unnotched beams with different length/depth ratio for a constant
out-of-plane thickness of 50mm were tested in order to study size
effect. The experimental tests were carried out applying a quasi-
static vertical loading at the top of the beams and by controlling

16
Vagg = 20%
14
Vagg = 30%
12
Vagg = 40%
Load (kN)

10
Vagg = 50%
8

0
0 0.05 0.1 0.15 0.2
Displacement (mm)
Fig. 11. Load-displacement curves for the specimens with different aggregate volume fractions.
120 E.A. Rodrigues et al. / International Journal of Solids and Structures 94–95 (2016) 112–124

16

14 Realization 1 Realization 2

12 Realization 3 Realization 4
10

Load (kN)
8

0
0 0.05 0.1 0.15 0.2
Displacement (mm)
Fig. 13. Load-displacement curves for the specimens with different random realizations of coarse aggregate.

Fig. 14. Geometry, boundary conditions and FE mesh of the beam B1.

Table 3
Material parameters for the beams.

Materials Young’s modulus Poisson’s ratio Fracture energy Tensile strength

Coarse aggregate Eagg = 50.0 GPa νagg = 0.2 – –

Mortar matrix Em = 30.2 GPa νm = 0 . 2 – –
Matrix–matrix interface Emi = 30.2 GPa νmi = 0 G fmi = 0.1 N/mm ftmi = 5.2 MPa
ITZ Eitz = 30.2 GPa νitz = 0 G fitz = 0.05 N/mm ftitz = 2.6 MPa

the crack mouth opening displacement (CMOD) at the bottom, as
illustrated in Fig. 14.
The so-called fifth-notched beams (with notch-to-depth ratio
of 0.2) are chosen to be numerically analyzed with the proposed
mesoscale approach. The beams of this group are classified ac-
cording to their depth as: B1 (D = 50 mm), B2 (D = 100 mm), B3
(D = 200 mm) and B4 (D = 400 mm), from smallest to largest beam
size. Fig. 14 shows the standard geometry, boundary conditions
and the FE mesh adopted for the beam B1. As can be seen in this
figure, the entire domain of the problem is discretized by generat-
ing computational models with 16,017; 27,174; 107,779 and 428,709
finite elements for the corresponding four beam sizes.
The aggregate particles are placed into the mortar matrix one
by one randomly, using the experimental particle size distribution
provided by Grégoire et al. (2013) and illustrated in Fig. 15, consid-
ering a volume fraction of 40% and varying the aggregate diameter
from dmin = 5 mm to dmax = 10 mm.
The material properties adopted for each constituent of the
mesoscale models are listed in Table 3. However, it is important to
mention that only Young’s modulus of concrete (Ec = 37GPa) was
provided by the experiments. Thereby, adopting a Young’s modulus
of aggregate equals to Eagg = 50 GPa, Young’s modulus of mortar
matrix was predicted using an inverse analysis based on the rule
of mixture, as proposed by Counto (1964). Then, the same value
was adopted for Young’s modulus of the interface elements.
The tensile strength of the matrix–matrix interface elements
and for the interface elements of the ITZ were calculated, so
that the average value is equivalent to that obtained experimen-
tally by Grégoire et al. (2013) for the concrete (mean value of
3.9 MPa obtained by the splitting tensile test of standard cylin-
ders experimentally analyzed), with a ratio between them of
2, i.e. ftmi / ftitz = 2. In addition, the same strategy were applied
for the fracture energy parameters, G fmi + G fitz /2 = GC , keeping
G fmi /G fitz = 2, where GC is the concrete fracture energy and its as-
sumed value is 0.075 N/mm, similar to that chosen numerically by
Grassl et al. (2012) (see Table 3).
The numerical analyses are carried out applying an incremental
vertical load at the top mid-point of the beams using the displace-
ment control method, controlling the vertical displacement of the
point where the load is applied.
Fig. 16 shows the crack pattern obtained for the four beams.
As expected, the strains are practically all concentrated in those
 E.A. Rodrigues et al. / International Journal of Solids and Structures 94–95 (2016) 112–124 121

100

Average sieve passage (%)

80

60

40

20

0
0 5 10 15 20 25 30
Average grain size (mm)
Fig. 15. Experimental grading curve of aggregate provided by Grégoire et al. (2013).

Fig. 16. Numerical crack patterns obtained for the beams: (a) B1, (b) B2, (c) B3 and (d) B4.
122 E.A. Rodrigues et al. / International Journal of Solids and Structures 94–95 (2016) 112–124
14
12
10
Force (kN)
8 Experimental D=200mm
6
4 Numerical D=400mm
0
0 0.1
Fig. 17. Comparison between numerical and experimental results: Force-CMOD curves.
interface elements near of the central cross section, following a
dominant vertical crack propagation (in mode I) from the notch,
with a zigzag path induced by the weakest ITZ and imposed by
the FE mesh of the mortar matrix.
Note that, although the crack patterns are very similar for the
four cases analyzed, the aggregate distribution directly influences
the crack path. The same observation was reported by other refer-
ences (Grégoire et al., 2013; Hordijk, 1992; Zhang et al., 2010). It is
also important to note that the geometry of the notch was defined
before the generation of the aggregates (see Fig. 16), which would
be equivalent to a cast-in notch. The results would have been more
realistic if the notch had gone through the aggregates.
The numerical load-CMOD curves are compared with those
obtained experimentally by Grégoire et al. (2013). As shown in
Fig. 17, the numerical and experimental results are in good agree-
ment for the beams B1, B2 and B3, while the numerical result
obtained for the beam B4, underestimates the peak-load and the
softening branch. However, it is important to note in the graphs
that the numerical model is capable of representing the size effect.
The results present the same tendency described in previous works
(Grassl et al., 2012; Grégoire et al., 2013; Skarżyński and Tejchman,
2010), in which the peak load increases and the softening branches
become more pronounced as the beam size increases.
6. Summary and conclusions
This paper presents an alternative 2D mesoscale model, in
which the concrete is considered as a heterogeneous three-phase
material composed of coarse aggregates, mortar matrix and ITZ.
The main novelty of the proposed approach is the use of interface
solid finite elements developed by Manzoli et al. (2012) to repre-
sent the complex ITZ and crack propagation process based on a
mesh fragmentation technique (Manzoli et al., 2016).
The shape of the coarse aggregates was represented by regu-
lar octagons and an efficient “take-and-place” method, proposed by
Wriggers and Moftah (2006), was used to generate and randomly
place the aggregates within the sample. This technique is very ap-
pealing since the aggregate size can be generated from an grading
curve. Studies including irregular aggregate could lead to interest-
ing results and this issue will be addressed in a future paper.
First, several uniaxial tension tests were numerically analyzed
in order to study the effects of aggregate parameters, such as size,
volume fraction and its distribution, on the tensile strength and
dissipated energy. In general, the mechanical response and crack
patterns were shown to be sensitive to the variation of these pa-
rameters. By increasing the maximum aggregate size, the peak load
Experimental D=50mm
Numerical D=50mm
Experimental D=100mm
Numerical D=100mm
Numerical D=200mm
Experimental D=400mm
0.2 0.3 0.4 0.5 0.6 0.7
CMOD (mm)
is reduced and the dissipated energy decreases. When the volume
fraction increases the peak load also decreases and, regarding the
load x displacement curves, the results were similar until the peak
load, followed by different softening branches that tend to con-
verge to a fixed residual strength.
Then, to investigate the capability of the strategy to simulate
structural members, three-point bending beams were numerically
analyzed and the results were compared with the experimental
ones obtained by Grégoire et al. (2013). The results demonstrated
that the mesoscale model was able to predict the size effect phe-
nomenon. Moreover, for the four geometrically similar notched
beams, the numerical results were in good agreement with those
obtained experimentally.
The results showed that the proposed model presents the same
kinds of characteristics that real 3D concrete shows, predicting the
macroscopic response of concrete, considering the effects of coarse
aggregate, mortar matrix and ITZ individually. It is important to
note that the strategy was formulated integrally in the framework
of the continuum mechanics and multiple arbitrary cracks could
be simulated without the need of a tracking algorithm. In addi-
tion, the introduction of this technique to existing FE code is very
simple, since only a few implementations in the pre-process stage,
related to the mesh fragmentation technique, are necessary.
In future works the authors intend to extend the formulation
presented here to 3D, in which the main challenge is the compu-
tational effort necessary to solve this type of problem, since the
3D mesh fragmentation technique could increase dramatically the
number of finite elements. Thus, multiscale analysis based on
the use of coupling finite elements (Bitencourt Jr. et al., 2015) will
also be addressed.
Acknowledgments
The authors wish to acknowledge the financial support of
the National Council for Scientific and Technological Development
(CNPq), Coordination for the Improvement of Higher Education
Personnel (CAPES) and Sao Paulo Research Foundation (FAPESP).
References
Bazant, Z.P., Tabbara, M.R., Kazemi, M.T., Pijaudier-Cabot, G., 1990. Random particle
model for fracture of aggregate or fiber composites. J. Eng. Mech. 116 (8), 1686–
1705. doi:10.1061/(ASCE)0733-9399(1990)116:8(1686).
Bitencourt Jr., L.A.G., Manzoli, O.L., Prazeres, P.G.C., Rodrigues, E.A., Bittencourt, T.N.,
2015. A coupling technique for non-matching finite element meshes. Comput.
Methods Appl. Mech. Eng. 290, 19–44. http://dx.doi.org/10.1016/j.cma.2015.02.
025. http://www.sciencedirect.com/science/article/pii/S0 0457825150 0 0870.
 E.A. Rodrigues et al. / International Journal of Solids and Structures 94–95 (2016) 112–124
Bolander Jr., J.E., Saito, S., 1998. Fracture analyses using spring networks with
random geometry. Eng. Fract. Mech. 61 (5-6), 569–591. http://dx.doi.org/
10.1016/S0013-7944(98)00069-1. http://www.sciencedirect.com/science/article/
pii/S0013794498000691.
Caballero, A., Carol, I., López, C., 2007. 3D meso-mechanical analysis of concrete
specimens under biaxial loading. Fatigue Fract. Eng. Mater. Struct. 30 (9), 877–
886. doi:10.1111/j.1460-2695.2007.01161.x.
Caballero, A., López, C., Carol, I., 2006. 3D meso-structural analysis of concrete
specimens under uniaxial tension. Comput. Methods Appl. Mech. Eng. 195
(52), 7182–7195. http://dx.doi.org/10.1016/j.cma.2005.05.052.Computational
Modelling of Concrete, http://www.sciencedirect.com/science/article/pii/
S0 0457825050 04068.
Carol, I., López, C., Roa, O., 2001. Micromechanical analysis of quasi-brittle materials
using fracture-based interface elements. Int. J. Numer. Methods Eng. 52 (1-2),
193–215. doi:10.1002/nme.277.
Carpinteri, A., Cornetti, P., Puzzi, S., 2004. A stereological analysis of aggregate grad-
ing and size effect on concrete tensile strength. Int. J. Fract. 128 (1–4), 233–242.
doi:10.1023/B:FRAC.0 0 0 0 040986.0 0333.86.
Counto, U.J., 1964. The effect of the elastic modulus of the aggregate on the elas-
tic modulus, creep and creep recovery of concrete. Mag. Concr. Res. 16 (48),
129–138.
Cusatis, G., Bažant, Z., Cedolin, L., 2006. Confinement-shear lattice CSL model
for fracture propagation in concrete. Comput. Methods Appl. Mech. Eng. 195
(52), 7154–7171. http://dx.doi.org/10.1016/j.cma.2005.04.019.Computational
Modelling of Concrete. http://www.sciencedirect.com/science/article/pii/
S0 0457825050 03956.
Cusatis, G., Mencarelli, A., Pelessone, D., Baylot, J., 2011. Lattice discrete par-
ticle model (LDPM) for failure behavior of concrete. II: Calibration and
validation. Cem. Concr. Compos. 33 (9), 891–905. http://dx.doi.org/10.1016/
j.cemconcomp.2011.02.010. http://www.sciencedirect.com/science/article/pii/
S09589465110 0 0333.
D’Addetta, G., Ramm, E., 2006. A microstructure-based simulation environment on
the basis of an interface enhanced particle model. Granul. Matter 8 (3-4), 159–
174. doi:10.1007/s10035- 006- 0004- 4.
De Schutter, G., Taerwe, L., 1993. Random particle model for concrete based on De-
launay triangulation. Mater. Struct. 26 (2), 67–73. doi:10.1007/BF02472853.
Du, X., Jin, L., Ma, G., 2014. Numerical modeling tensile failure behavior of concrete
at mesoscale using extended finite element method. Int. J. Damage Mech. 23
(7), 872–898. http://dx.doi.org/10.1177/1056789513516028.
Eliáš, J., Stang, H., 2012. Lattice modeling of aggregate interlocking in concrete. Int.
J. Fract. 175 (1), 1–11. doi:10.1007/s10704- 012- 9677- 3.
Garboczi, E., 2002. Three-dimensional mathematical analysis of particle shape
using X-ray tomography and spherical harmonics: Application to aggregates
used in concrete. Cem. Concr. Res. 32 (10), 1621–1638. http://dx.doi.org/
10.1016/S0 0 08-8846(02)0 0836-0. http://www.sciencedirect.com/science/article/
pii/S0 0 088846020 08360.
Grassl, P., Grégoire, D., Solano, L.R., Pijaudier-Cabot, G., 2012. Meso-scale modelling
of the size effect on the fracture process zone of concrete. Int. J. Solids Struct.
49 (13), 1818–1827. http://dx.doi.org/10.1016/j.ijsolstr.2012.03.023. http://www.
sciencedirect.com/science/article/pii/S0020768312001175.
Grassl, P., Jirásek, M., 2010. Meso-scale approach to modelling the fracture pro-
cess zone of concrete subjected to uniaxial tension. Int. J. Solids and Struct.
47 (7-8), 957–968. http://dx.doi.org/10.1016/j.ijsolstr.2009.12.010. http://www.
sciencedirect.com/science/article/pii/S0 0207683090 04752.
Grassl, P., Pearce, C., 2010. Mesoscale approach to modeling concrete subjected to
thermomechanical loading. J. Eng. Mech. 136 (3), 322–328. doi:10.1061/(ASCE)
0733-9399(2010)136:3(322).
Grégoire, D., Rojas-Solano, L., Pijaudier-Cabot, G., 2013. Failure and size effect for
notched and unnotched concrete beams. Int. J. Numer. Anal. Methods Geomech.
37 (10), 1434–1452. doi:10.1002/nag.2180.
Häfner, S., Eckardt, S., Luther, T., Könke, C., 2006. Mesoscale modeling of concrete:
Geometry and numerics. Comput. Struct. 84 (7), 450–461. http://dx.doi.org/10.
1016/j.compstruc.2005.10.003. http://www.sciencedirect.com/science/article/pii/
S0 0457949050 03585.
Hordijk, D., 1992. Tensile and tensile fatigue behaviour of concrete; experiments,
modeling and analyses. Heron 37 (1), 1–79.
Huang, Y., Yang, Z., Ren, W., Liu, G., Zhang, C., 2015. 3D meso-scale frac-
ture modelling and validation of concrete based on in-situ X-ray com-
puted tomography images using damage plasticity model. Int. J. Solids
Struct. 67-68, 340–352. http://dx.doi.org/10.1016/j.ijsolstr.2015.05.002. http://
www.sciencedirect.com/science/article/pii/S0 0207683150 02127.
Kim, S.-M., Al-Rub, R.K.A., 2011. Meso-scale computational modeling of the plastic-
damage response of cementitious composites. Cem. Concr. Res. 41 (3), 339–358.
http://dx.doi.org/10.1016/j.cemconres.2010.12.002.
Landis, E., Bolander, J., 2009. Explicit representation of physical processes in
concrete fracture. J. Phys. D Appl. Phys. 42 (21), 1–17. http://stacks.iop.org/
0022-3727/42/i=21/a=214002.
Leite, J., Slowik, V., Mihashi, H., 2004. Computer simulation of fracture processes
of concrete using mesolevel models of lattice structures. Cem. Concr. Res. 34
(6), 1025–1033. http://dx.doi.org/10.1016/j.cemconres.2003.11.011. http://www.
sciencedirect.com/science/article/pii/S0 0 088846030 04083.
Lilliu, G., van Mier, J., 2003. 3D lattice type fracture model for concrete.
Eng. Fract. Mech. 70 (7-8), 927–941. http://dx.doi.org/10.1016/S0013-7944(02)
0 0158-3. http://www.sciencedirect.com/science/article/pii/S0 0137944020 01583.
123
López, C.M., Carol, I., Aguado, A., 2008. Meso-structural study of concrete fracture
using interface elements. I: numerical model and tensile behavior. Mater. Struct.
41 (3), 583–599. doi:10.1617/s11527- 007- 9314- 1.
Manzoli, O., Gamino, A., Rodrigues, E., Claro, G., 2012. Modeling of interfaces in
two-dimensional problems using solid finite elements with high aspect ra-
tio. Comput. Struct. 94-95 (0), 70–82. doi:10.1016/j.compstruc.2011.12.001. http:
//www.sciencedirect.com/science/article/pii/S0 0457949110 02938.
Manzoli, O.L., Maedo, M.A., Bitencourt Jr., L.A., Rodrigues, E.A., 2016.
On the use of finite elements with a high aspect ratio for model-
ing cracks in quasi-brittle materials. Eng. Fract. Mech. 153, 151–170.
http://dx.doi.org/10.1016/j.engfracmech.2015.12.026. http://www.sciencedirect.
com/science/article/pii/S0 0137944150 0702X.
van Mier, J., 1991. Mode I fracture of concrete: Discontinuous crack growth and
crack interface grain bridging. Cem. Concr. Res. 21 (1), 1–15. http://dx.doi.org/10.
1016/0 0 08-8846(91)90 025-D. http://www.sciencedirect.com/science/article/pii/
0 0 0888469190 025D.
van Mier, J., 1997. Fracture Processes of Concrete: Assessment of Material Parame-
ters for Fracture Models. CRC Press.
van Mier, J.G., Nooru-Mohamed, M., 1990. Geometrical and structural as-
pects of concrete fracture. Eng. Fract. Mech. 35 (4-5), 617–628. http://dx.
doi.org/10.1016/0013-7944(90)90144-6. http://www.sciencedirect.com/science/
article/pii/0013794490901446.
Mier, J.V., Vliet, M.V., 2003. Influence of microstructure of concrete on
size/scale effects in tensile fracture. Eng. Fract. Mech. 70 (16), 2281–
2306. http://dx.doi.org/10.1016/S0013- 7944(02)00222- 9.Size-scale effects, http:
//www.sciencedirect.com/science/article/pii/S0 0137944020 02229.
Monteiro, P., Maso, J., Ollivier, J., 1985. The aggregate-mortar interface. Cem. Concr.
Res. 15 (6), 953–958. http://dx.doi.org/10.1016/0 0 08-8846(85)90 084-5. http://
www.sciencedirect.com/science/article/pii/0 0 0888468590 0845.
Niknezhad, D., Raghavan, B., Bernard, F., Kamali-Bernard, S., 2015. To-
wards a realistic morphological model for the meso-scale mechani-
cal and transport behavior of cementitious composites. Compos. Part
B Eng. 81, 72–83. http://dx.doi.org/10.1016/j.compositesb.2015.06.024.
http://www.sciencedirect.com/science/article/pii/S13598368150 040 0X.
Oliver, J., Caicedo, M., Roubin, E., Huespe, A., Hernández, J., 2015. Continuum ap-
proach to computational multiscale modeling of propagating fracture. Comput.
Methods Appl. Mech. Eng. 294, 384–427. http://dx.doi.org/10.1016/j.cma.2015.05.
012. http://www.sciencedirect.com/science/article/pii/S0 0457825150 01851.
Oliver, J., Huespe, A., Blanco, S., Linero, D., 2006. Stability and robustness issues in
numerical modeling of material failure with the strong discontinuity approach.
Comput. Methods Appl. Mech. Eng. 195 (52), 7093–7114.Computational Mod-
elling of Concrete.
Oliver, J., Huespe, A., Cante, J., 2008. An implicit/explicit integration scheme to
increase computability of non-linear material and contact/friction problems.
Comput. Methods Appl. Mech. Eng. 197 (21–24), 1865–1889. http://dx.doi.
org/10.1016/j.cma.2007.11.027. http://www.sciencedirect.com/science/article/pii/
S0 0457825070 04756.
Prazeres, P.G.C., Bitencourt Jr., L.A.G., Bittencourt, T.N., Manzoli, O.L., 2016. A
modified implicit–explicit integration scheme: an application to elastoplas-
ticity problems. J. Braz. Soc. Mech. Sci. Eng. 38 (1), 151–161. doi:10.1007/
s40430- 015- 0343- 3.
Rodrigues, E.A., Manzoli, O.L., Bitencourt Jr., L.A.G., Prazeres, P.G.C., Bittencourt, T.N.,
2015. Failure behavior modeling of slender reinforced concrete columns sub-
jected to eccentric load. Lat. Am. J. Solids Struct. 12 (3), 520–541. http://dx.doi.
org/10.1590/1679-78251224.
Roubin, E., Vallade, A., Benkemoun, N., Colliat, J.-B., 2015. Multi-scale fail-
ure of heterogeneous materials: A double kinematics enhancement for em-
bedded finite element method. Int. J. Solids Struct. 52, 180–196. http:
//dx.doi.org/10.1016/j.ijsolstr.2014.10.001. http://www.sciencedirect.com/science/
article/pii/S0 0207683140 03710.
Sánchez, M., Manzoli, O., Guimarães, L., 2014. Modeling 3-D desiccation soil crack
networks using a mesh fragmentation technique. Comput. Geotech. 62 (0), 27–
39. http://dx.doi.org/10.1016/j.compgeo.2014.06.009. http://www.sciencedirect.
com/science/article/pii/S0266352X14001177.
Schlangen, E., Garboczi, E., 1997. Fracture simulations of concrete using lattice
models: Computational aspects. Eng. Fract. Mech. 57 (2-3), 319–332. http://dx.
doi.org/10.1016/S0013-7944(97)00010-6. http://www.sciencedirect.com/science/
article/pii/S0 0137944970 0 0106.
Schlangen, E., van Mier, J., 1992. Simple lattice model for numerical simulation of
fracture of concrete materials and structures. Mater. Struct. 25 (9), 534–542.
doi:10.1007/BF02472449.
Skarżyński, Ł., Tejchman, J., 2010. Calculations of fracture process zones on meso-
scale in notched concrete beams subjected to three-point bending. Eur. J. Mech.
A Solids 29 (4), 746–760. http://dx.doi.org/10.1016/j.euromechsol.2010.02.008.
http://www.sciencedirect.com/science/article/pii/S0997753810 0 0 0379.
Tejchman, J., Bobiński, J., 2013. Continuous and Discontinuous Modelling of Fracture
in Concrete using FEM. Springer Series in Geomechanics and Geoengineering,
Springer. Berlin doi:10.1007/978- 3- 642- 28463- 2.
Unger, J.F., Eckardt, S., 2011. Multiscale modeling of concrete: From mesoscale
to macroscale. Arch. Comput. Methods Eng. 18 (3), 341–393. doi:10.1007/
s11831-011-9063-8.
Wang, X., Yang, Z., Jivkov, A.P., 2015. Monte Carlo simulations of mesoscale fracture
of concrete with random aggregates and pores: a size effect study. Constr. Build.
Mater. 80, 262–272. http://dx.doi.org/10.1016/j.conbuildmat.2015.02.002. http://
www.sciencedirect.com/science/article/pii/S0950 0618150 01154.
124 E.A. Rodrigues et al. / International Journal of Solids and Structures 94–95 (2016) 112–124
Wang, Z., Kwan, A., Chan, H., 1999. Mesoscopic study of concrete I: Genera-
tion of random aggregate structure and finite element mesh. Comput. Struct.
70 (5), 533–544. http://dx.doi.org/10.1016/S0045-7949(98)00177-1. http://www.
sciencedirect.com/science/article/pii/S0045794998001771.
Wittmann, F.H., Sadouki, H., Steiger, T., 1993. Experimental and numerical
study of effective properties of composite materials. Huet, C., Microme-
chanics of Concrete and Cementitious Composites, Universitaires Romandes
Lausanne.
de Wolski, S., Bolander, J., Landis, E., 2014. An in-situ X-ray microtomography study
of split cylinder fracture in cement-based materials. Exp. Mech. 54 (7), 1227–
1235. doi:10.1007/s11340- 014- 9875- 1.
Wriggers, P., Moftah, S., 2006. Mesoscale models for concrete: Homogenisation
and damage behaviour. Finite Elem. Anal. Des. 42 (7), 623–636. http://dx.doi.
org/10.1016/j.finel.20 05.11.0 08.The Seventeenth Annual Robert J. Melosh Com-
petition. The Seventeenth Annual Robert J. Melosh Competition, http://www.
sciencedirect.com/science/article/pii/S0168874X05001563.
Zhang, J., Leung, C., Xu, S., 2010. Evaluation of fracture parameters of concrete from
bending test using inverse analysis approach. Mater. Struct. 43 (6), 857–874.
doi:10.1617/s11527- 009- 9552- 5.
Zubelewicz, A., Bažant, Z., 1987. Interface element modeling of fracture in ag-
gregate composites. J. Eng. Mech. 113 (11), 1619–1630. doi:10.1061/(ASCE)
0733-9399(1987)113:11(1619).